Expt1- Recognition Memory

Load in packages

Read in raw groilla data

# read in low test expect data exported from gorilla
setwd(here::here('SF_data', 'Gorilla_data_low'))

data=here::here('SF_data', 'Gorilla_data_low')  # path to data files

file_list=list.files(data, pattern=".csv") # list of data files
 
# read in all files
datasetlow <-
  do.call("rbind", lapply(file_list, FUN=function(files){
    
    for (i in 1:length(files)){ 
      if(file.exists(files[i])){
        message( "now processing:", files[i])
      }
    }
    fread(files, header=TRUE, sep=",", na.strings = "", fill=TRUE)})) #fread makes reading in files quick
#

library(lubridate)
# clean up data! Select data from after the pre-registation! 


low<-datasetlow %>% 
  janitor::clean_names(.) %>% 
  dplyr::mutate(date=as.Date(utc_date)) %>%
  dplyr::filter(date=="08/06/2020" |date=="09/06/2020" , zone_type=="response_button_text")

#response as character
low$response<-as.character(low$response)

#assign column to denot low test expect
low$testexpect<-"low"

# high test expect
setwd(here::here('SF_data', 'Gorilla_data_high'))

data=here::here('SF_data', 'Gorilla_data_high')  # path to data files

file_list=list.files(data, pattern=".csv") # list of data files
 
# read in all files
highdata <-
  do.call("rbind", lapply(file_list, FUN=function(files){
    
    for (i in 1:length(files)){ 
      if(file.exists(files[i])){
        message( "now processing:", files[i])
      }
    }
    fread(files, header=TRUE, sep=",", na.strings = "", fill=TRUE)})) #fread makes reading in files quick
#

library(lubridate)
# a batch of Ss we run before preregistration that should not be included in the analysis
high <-highdata %>% 
  janitor::clean_names(.) %>% 
  dplyr::mutate(date=as.Date(utc_date)) %>%
  dplyr::filter(date=="08/06/2020" | date=="0009/07/2020" |date=="0010/07/2020" | date=="09/06/2020", zone_type=="response_button_text")

#response as character
high$response<-as.character(high$response)

# assign column to denot high test expect
high$testexpect<-"high"

# bind low and high datasets
high_low<-rbind(high, low)

Recognition memory

Calculate d-prime

#response as character
#calculate hit rate and far and compute dprime and other measures
ex4=high_low %>% dplyr::mutate(condition1= dplyr::case_when( 
  condition == "SF" ~ "Sans Forgetica", 
  condition =="normal" ~  "Arial", 
), isold= dplyr::case_when (
  old_new== "old" ~ 1, 
  old_new== "new" ~ 0), 
sayold=dplyr::case_when( 
  response=="old"~ 1, 
  response=="new" ~ 0, 
  ))


#classic SDT
sdt <- ex4 %>% 
  dplyr::mutate(type = "hit",
         type = ifelse(isold==1 & sayold==0, "miss", type),
         type = ifelse(isold==0 & sayold==0, "cr", type),  # Correct rejection
         type = ifelse(isold==0 & sayold==1, "fa", type))  # False alarm
sdt <- sdt %>% 
  dplyr::group_by(participant_private_id, type, condition1, testexpect) %>% 
  dplyr::summarise(count = n()) %>% 
  tidyr::spread(type, count)  # Format data to one row per person

sdt <- sdt %>% 
  dplyr::group_by(participant_private_id, condition1, testexpect)%>%
  dplyr::mutate(hr = hit / (hit+miss),
         fa = fa / (fa+cr)) %>%
  dplyr::mutate(hr=case_when(
    is.na(hr) ~ 0.99,
    TRUE ~ hr), 
    fa=case_when(
      is.na(fa)  ~ 0.01,
    TRUE ~ fa),
     zhr=qnorm(hr), 
     zfa=qnorm(fa), 
    dprime = zhr-zfa) %>%
  ungroup()

Plot

Raincloud plots (Allen et al., 2019) depicting raw data (dots), box plots, and half violin kernel desntiy plots, with mean (red dot). Proportion of “old” responses as a function of Test Expectancy for Experiment 1.

#set up raincloud params
# fig for dprime

highlowaov=sdt  %>% select(participant_private_id, condition1, testexpect, dprime) %>%
  mutate(testexpect=ifelse(testexpect=="low", "Low Test Expectancy", "High Test Expectancy"))

#plot

fig1 <- ggplot(highlowaov,aes(x=condition1,y=dprime,fill=condition1))+ facet_grid(~testexpect) + 
  geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) + 
  geom_boxplot(aes(x = condition1, y = dprime),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
  stat_summary(fun=mean, geom="point", colour="darkred", size=3) + 
  theme_cowplot() +
  scale_colour_brewer(palette = "Dark2")+
  scale_fill_brewer(palette = "Dark2") +
  labs(y = "Sensitivity(d')", x = "Typeface") + theme(legend.position = "none")

fig1

ggsave("dprimefig1.png", width=8, height=4)

Analysis

#ANOVA


a1 <- aov_ez("participant_private_id", "dprime", highlowaov, 
             between = c("testexpect"), within=c("condition1")) # mixed

summary(a1)

## 
## Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
## 
##                        Sum Sq num Df Error SS den Df  F value    Pr(>F)    
## (Intercept)           296.652      1  166.184    229 408.7834 < 2.2e-16 ***
## testexpect              2.980      1  166.184    229   4.1058  0.043896 *  
## condition1              1.818      1   38.786    229  10.7344  0.001215 ** 
## testexpect:condition1   0.735      1   38.786    229   4.3369  0.038405 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

#kable(summary(a1))

JOLs

Read in JOL data

# get JOls from raw data

high

jol_high<- highdata %>% 
  mutate(testexpect="high")
low

jol_low<-datasetlow %>%
  mutate(testexpect="low")
jol_high_low <- rbind(jol_high, jol_low)

#bind high and low
jols<-jol_high_low %>% janitor::clean_names(.) %>% dplyr::mutate(date=as.Date(utc_date)) %>%  dplyr::filter(date=="08/06/2020" | date=="0009/07/2020"|date=="0010/07/2020" | date=="09/06/2020",  zone_type=="response_slider_endValue" | zone_type=="response_text_entry")


jols$response<-as.numeric(jols$response)


jols1<- jols %>%
  dplyr::select(participant_private_id, response, testexpect) %>%
  dplyr::mutate(cond=rep(1:2, 231), font=ifelse(cond==1, "SF", "A")) %>%
  tidyr::drop_na() %>% 
  dplyr::mutate(testexpect=ifelse(testexpect=="low", "Low Test Expectancy", "High Test Expectancy"), font=ifelse(font=="A", "Arial", "Sans Forgetica"))

Plot JOls

# plot JOLs

figjol <- ggplot(jols1,aes(x=font,y=response,fill=font))+ facet_grid(~testexpect) + 
  geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) + 
  geom_boxplot(aes(x = font, y = response),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
    stat_summary(fun=mean, geom="point", colour="darkred", size=5)+
  theme_cowplot() +
  scale_colour_brewer(palette = "Dark2")+
  scale_fill_brewer(palette = "Dark2") +
  labs(y = "Judgements of Learning", x = "Typeface") + theme(legend.position = "none")

ggsave("figjol.png", width=8, height=4, dpi=300)

figjol

#6.67

Analysis

#anova JOLs
a1 <- aov_ez("participant_private_id", "response", jols1, 
             between = c("testexpect"), within=c("font")) # mixed

summary(a1)

## 
## Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
## 
##                  Sum Sq num Df Error SS den Df  F value    Pr(>F)    
## (Intercept)     1575566      1   403702    221 862.5179 < 2.2e-16 ***
## testexpect        29254      1   403702    221  16.0145 8.579e-05 ***
## font               1803      1    14729    221  27.0520 4.509e-07 ***
## testexpect:font       9      1    14729    221   0.1334    0.7152    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

RTs

Read in RT data

#raw gorilla data and extract RTs

datasetlow$testexpt<-"low"
highdata$testexpt<-"high"

rt_high_low <- rbind(datasetlow, highdata)

rt<-rt_high_low %>% janitor::clean_names(.) %>% mutate(date=as.Date(utc_date)) %>%  dplyr::filter(date=="08/06/2020" | date=="0009/07/2020"|date=="0010/07/2020" | date=="09/06/2020", zone_type=="continue_button", display=="study") 

# get RT and make numeric (Gorilla does not do this)
rt$reaction_time<-as.numeric(rt$reaction_time)

rt1<- rt %>% 
  dplyr::group_by(participant_private_id, condition, testexpt) %>% 
  dplyr::select(participant_private_id, condition, testexpt, reaction_time) %>%
  dplyr::mutate(sdabove = mean(reaction_time, na.rm=TRUE) +  2.5*sd(reaction_time, na.rm=TRUE)) %>%
    dplyr::filter(reaction_time > 150, reaction_time < sdabove) %>%
  dplyr::summarise(mean_rt= mean(log(reaction_time))) %>%
   mutate(testexpt=ifelse(testexpt=="low", "Low Test Expectancy", "High Test Expectancy"), font=ifelse(condition=="normal", "Arial", "Sans Forgetica")) %>%
  select(-condition)

Plot RT data

#plot fig

figrt <- ggplot(rt1,aes(x=font,y=mean_rt,fill=condition))+ facet_grid(~testexpt) + 
  geom_flat_violin(position = position_nudge(x = .2, y = 0), alpha = .4,adjust=4)+
  geom_point(position=position_jitter(width = .15),size = 1, alpha = 0.2) + 
  geom_boxplot(aes(x = font, y = mean_rt),outlier.shape = NA,
               alpha = 0.3, width = .1, colour = "BLACK") +
    stat_summary(fun=mean, geom="point", colour="darkred", size=5)+
  theme_cowplot() +
  scale_colour_brewer(palette = "Dark2")+
  scale_fill_brewer(palette = "Dark2") +
  labs(y = "log(Study Times/ms)", x = "Typeface") + theme(legend.position = "none")


ggsave("figrt.png", width=8, height=4, dpi=300)

figrt

#write.csv(rt2, file="rt_high_low.csv")

#ttestBF(x=rt2$normal, y=rt

Analysis

#anova RTs

a1 <- aov_ez("participant_private_id", "mean_rt", rt1, 
             between = c("testexpt"), within=c("condition")) # 

summary(a1)

## 
## Univariate Type III Repeated-Measures ANOVA Assuming Sphericity
## 
##                     Sum Sq num Df Error SS den Df    F value    Pr(>F)    
## (Intercept)        20706.2      1  168.431    229 28152.2648 < 2.2e-16 ***
## testexpt               1.1      1  168.431    229     1.5354    0.2166    
## condition              0.3      1    1.797    229    33.0251 2.884e-08 ***
## testexpt:condition     0.0      1    1.797    229     1.1292    0.2891    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

#kable(summary(a1))